Node Index

Every node in winkComposer, grouped by category. Nodes that serve a secondary purpose carry a #Tag marking the additional category.

Arithmetic

Simple math operations between message fields.

accumulateRunning total of a field

diffDifference between two fields

invertFlagInvert a boolean field

ratioDivide one field by another

transformApply a pure function to each sample

Detection

Spot anomalies, confirm them, and measure process capability.

pageHinkleyDetect mean shifts (change-point)

persistenceCheckConfirm condition holds for N of M samples

processIndexCompute Cpk/Ppk capability index#FeatureExtraction

thresholdCheck if value crosses a boundary

Feature Extraction

Extract patterns from streaming data — trends, correlations, state changes, and windowed statistics.

digestMomentsExpand compact digest into full statistics

dwellTimeTrackerTrack how long a condition has been active#Detection

esCorrelationExponentially smoothed Pearson correlation

esPairwiseCorrelationAll pairwise correlations across a field group

lagCompare current value with lagged value

momentsDigestCompact statistical digest over a window

stateChangeDetectorDetect state changes with debounce#Detection

trendClassify signal as stable, rising, or falling#Detection

vectorDistanceDistance between two vectors

Flow Control

Gate messages based on conditions.

passIfDrop messages that fail a condition

Intelligence

Model-based reasoning — fuse evidence, estimate hidden state, and detect anomalies that simpler nodes miss.

appraiseFuse evidence from multiple signals into a health conviction score#Detection

kalman1dModel-based state estimation with outlier detection#SignalConditioning

Observability

Broadcast insights and persist data to external systems.

emitIfBroadcast to MQTT when condition is met

persistIfWrite to storage when condition is met

Orchestration

Coordinate other nodes dynamically — enable, disable, reset, pause, and flush on conditions.

controllerEnable, disable, reset, pause, and flush nodes on conditions

Signal Conditioning

Clean and prepare raw data before it reaches the rest of the pipeline.

butterworthFilter2nd-order low/high-pass filter

categorizeMap values to named categories#FeatureExtraction

esMeanExponentially smoothed mean

esStatsSmoothed mean, stdev, envelope, zScore#FeatureExtraction

kernelConvolution: smoothing, derivatives, detection#Detection

median33-sample median filter

sanitizeValidate inputs against ranges and rules#Detection

spikeGuardDetect and clean isolated spikes#Detection

swStatsSliding-window exact statistics

twStatsTumbling-window statistics#FeatureExtraction

Next Steps

Flow Language — how to use these nodes in a pipeline
Composition Patterns — proven combinations for common problems